Nested For Loop for 2D Matrix

I want to create a nested for loop that essentially goes through a 2D matrix of 600x1000 dimensions, such that in every loop in both of the dimensions it picks out exactly 51x51 part of the 2D matrix in either of the dimensions and multiplies it to another 51x51 separate matrix.

Answer 1

[Matt J]

So then you want to extract random 51x51 submatrices? If so, then,

I=sort(randperm(600,51));
J=sort(randperm(1000,51));

submatrix= matrix(I,J);

[BBB]

Sorry if the question was not very clear. I have a data set which is 2D 600x1000 in dimensions. I have to create a nested for loop of the following format,

for ii = 1:1000

for jj = 1:600

end

end

....

which will go through the 1000 dim first taking the first 51x51 elements (the first column should be col 1-51 in the first loop, the second iteration sould be 2-52, third 3-53--as long as it reaches till the end). For the 600 dim as well, taking the first 51x51 elements (the first row should be 1-51 in the first loop, the second iteration sould be 2-52, third 3-53--as long as it reaches till the end). Each of these set of 51x51 should simultaneously be multiplied with a static set matrix that I have which is also 51x51.

[Matt J]

Is it element-wise multiplication, or another kind? And what is to be done with each of the 51x51 multiplication results? In double floating point, this will require 10 GB of storage if they are kept simultaneously in memory,

>> (600-51)*(1000-51)*51^2*8/2^30

ans =

   10.0965

Answer 2

[Matt J]

I vaguely wonder whether you are just re-inventing convolution

result = conv2(Matrix,staticMatrix,'valid');